How are the HMM cutoff scores in TIGRFAMs determined?

Question

[Rui]

Hello everyone,

As indicated in the title, I am wondering how are the HMM cutoff scores in TIGRFAMs determined?

Where did I find the score? (If you search a TIGRFAM accession in the NCBI protein family model database, e.g. TIGR02064.1, you will be directed to another page that displays information about this HMM including the NCBI HMM accession, GO terms, HMM length, sequence cutoff, domain cutoff, etc.)

The cutoffs I was talking about are the sequence cutoff and domain cutoff, which are usually equal. You can then specify this score in the hmmsearch program if you want to search for a protein.

Thank you!

Rui

Answer 1

[Mensur Dlakic]

These scores are set manually by those who made the alignments and created HMMs. This is not something that all HMMs have - it has to be set on purpose during the HMM building process.

It is usually done to catch true homologs that have E-values larger than the inclusion threshold, as HMMs would otherwise miss them. If I remember correctly, it used to be done by scoring all known true homologs, and then picking the lowest bit-score of the known true homolog.

[Rui]

Thanks for your answer!